Direct entry of cell-penetrating peptide can be controlled by maneuvering the membrane curvature

A biomembrane's role is to be a barrier for interior cytosol from an exterior environment to execute the cell's normal biological functions. However, a water-soluble peptide called cell-penetrating peptide (CPP) has been known for its ability to directly penetrate through the biomembranes into cells (cytolysis) without perturbating cell viability and expected to be a promising drug delivery vector. Examples of CPP include peptides with multiple arginine units with strong cationic properties, which is the key to cytolysis. Here we show the conclusive evidence to support the mechanism of CPP’s cytolysis and way to control it. The mechanism we proposed is attributed to biomembrane’s physicochemical nature as lamellar liquid crystal (Lα). Cytolysis occurs as the temporal and local dynamic phase transitions from Lα to an undulated lamellar with pores called Mesh1. We have shown this phase transfer of Lα composed of dioleoyl-phosphatidylcholine (DOPC) with water by adding oligo-arginine (Rx) as CPP at the equilibrium. Using giant unilamellar vesicle composed of DOPC as a single cell model, we could control the level of cytolysis of CPP (FITC-R8) by changing the curvature of the membrane through osmotic pressure modulation. The cytolysis of CPP utilizes biomembrane's inherent topological and functional flexibility corresponding to the stimuli.


Results and discussion
Effect of oligo-arginine to induce positive curvature for membrane lipid under equilibrium condition. In order to clarify whether CPP could make membrane curvature positive by reducing SP value, we have investigated the effect of oligo-arginine (Rx) on the phase structure of LC composed of dioleoyl-phos- www.nature.com/scientificreports/ phatidylcholine (DOPC) with water under the equilibrium condition by small-angle Xray scattering (SAXS) measurement as shown in Fig. 2. We can determine the LC structure from the SAXS peaks through the crystallographic analysis of peak ratios expressed as q values. Theoretically, LC structure would change from lamellar (Lα) to bicontinuous cubic (V 1 ) then hexagonal (H 1 ) by reducing SP value or increasing curvature toward the water, and Mesh 1 would exist in between Lα and V 1 (Fig. S1), and along with this phase transitions, peak ratio changes from 1:2:3 for Lα and Mesh 1, then 1:√6:√8 for V 1 , and 1:√3:2 for H 1 theoretically. As shown in Fig. 2a, q values of SXAS peaks for hexa-arginine (R6) added to the DOPC/Water (60/40 wt%) system followed 1:2:3 q-peak ratio as Lα within the range of DOPC: R6 molar ratio from 1:0.083 (a-2) to 1:0.15 (a-4) (Fig. 2). The V 1 phase's coexistence in Lα was indicated for (a-3) and (a-4) by the shoulder peaks shown by vertical dotted lines. Further incremental addition of R6 changed the structure to V 1 (1:√6:√8 ) (a-5) then to H 1 (1:√3:2) (a-6). We also made polarized microscope observations for each sample, as shown in Fig. 2b which agree with the SAXS peak analysis. Namely, oil-streak texture for the Lα phase (a-1 to a-4), no birefringence as a cubic phase for the V 1 (a-5), and fan-like textures for the H 1 (a-6). These consecutive phase changes by the addition of R6 are the proof of substantial cationic property of guanidium moiety of arginine in R6 to interact with phosphate anion of DOPC to make the hydrophilic part more considerable corresponding to the larger a value (decreasing SP in Eq. (1) and increasing membrane curvature. Fig. S1).
It is noteworthy that interlayer spacing D calculated by the first peak q value increased consecutively within the Lα state (DOPC: R6 up to 1:0.15), as shown in Fig. 2a and c. Usually, such an increasing D value is observed by increasing the water ratio resulting in the interlayer water thickness larger, as shown in Supplementary Fig. S3 for a simple DOPC/Water system without Rx. Namely, interlayer spacing D increased consecutively by decreasing DOPC concentration and reached a constant level of 6.4 nm (Fig. S3b,d) at DOPC < 50 wt% with the formation of vesicles confirmed by maltase cross texture as shown in (Fig. S3c). On the other hand, R6 has added to the DOPC/Water (60/40) system in the current system (Fig. 2). Namely, the current system's water amount is fixed so that water layer expansion in the planar lamellar layers would not be possible (Fig. 2c).
We observed similar D layer expansion in Lα when R4 or R8 was added to the DOPC/Water (60/40) system. As a strong cation, Rx induces positive curvature by increasing a value in Eq. (1) under equilibrium condition. Here we assume this D value increase is attributed to the curved deformation of planar Lα structure to undulated lamellar as shown in Fig. 2c, which reasonably fits the Mesh 1 structure as a metastable phase in between Lα and V 1 . Hyde et al. reported the Mesh 1 phase as a punctured bilayer to accommodate increased vending energy by the  Effect of osmotic pressure to control cytolysis. In order to establish a way to control CPP Cytolysis, we made the second hypothesis that CPP's translocation could be controlled if the positive curvature change is stimulated or suppressed utilizing physical or chemical modulation 9 . To prove this, we first used erythrocyte (red blood cell), which has a slightly negative mean membrane curvature and an ability to change its shape by osmotic pressure corresponding to the curvature modulation. We have applied osmotic pressure within the range of no hemolysis to keep the barrier function intact. We found that erythrocyte is resistant to CPP penetration compared to normal cells at isotonic osmotic pressure. Moreover, we could promote or suppress CPP's penetration to erythrocytes under hypotonic or hypertonic conditions, respectively 9,10,13 (Supplementary Fig. S4 10 ). As a result, we succeeded in controlling CPP cytolysis by osmotic pressure as a physical stimulus 9,10,13 . To our best knowledge, this is the first successful attempt to control the cytolysis of CPP by membrane curvature modulation. In order to confirm the second hypothesis further, we chose a giant unilamellar vesicle (GUV) as a model cell membrane. We have investigated the effect of osmotic pressure on the penetration of CPP to the GUV made of egg yolk phosphatidylcholine (EPC). GUV formation was confirmed by differential interference contrast microscope (DICM) observation. The size distribution was measured by dynamic light scattering (DLS), as shown in Supplementary Fig. S5.
Cytolysis of FITC-R8 to EPC GUV was conducted similarly to erythrocyte by changing the osmotic pressure, as shown in Fig. 3. The amount of cytolysis of FITC-R8 increased by reducing the osmotic pressure, corresponding to the curvature increase, which is in good agreement with the results obtained for erythrocyte [9][10][11]13 . The recovery of FITC-R8 from supernatant inversely decreased by osmotic pressure reduction (Fig. 3). Thus, this experiment's reliability was confirmed by the total recovery rate of applied FITC-R8, as shown in Supplementary  Table T1.
Under the tested conditions, FITC alone could not permeate through the membrane into the cytosol regardless of osmotic pressure change 10 . The GUV membrane's barrier function remained intact as the leakage of rhodamine pre-encapsulated in the GUV stayed very low and consistent under different osmotic pressures neither in the presence nor absence of FITC-R8 in the outer medium (Fig. 4). Furthermore, pre-internalized FITC-R8 could not get out of the GUV regardless of the osmotic pressure as shown in (Fig. 4). These results reasonably support the two hypotheses since the inner lipid membrane has negative curvature, preventing phase transition to Mesh 1 .
Next, we have confirmed that osmotic pressure-dependent cytolysis of CPP is typical for GUVs composed of different varieties of phospholipids such as EPC, Dioleoyl-phosphatidylcholine (DOPC), and palmitoyl-oleoylphosphatidylcholine (POPC) as shown in Fig. 5. The cytolysis level is higher for EPC which is natural lecithin, as a mixture of various fatty acid residues. In terms of the effect of temperature, the penetration efficiency of FITC-R8 to DOPC was slightly low at 2 °C with statistical significance compared with that at 37 °C at all osmotic pressures (Fig. 6). This result indicates that membrane flexibility affects the FITC-R8 permeability, as we have previously reported. Namely, we found no cytolysis of FITC-R8 to SOPC GUV at 0 °C under confocal laser scanning microscope observation 11 . This is because SOPC has Tc at 6 °C and SOPC GUV is under gel state as hydrated solid at 0 °C.
These results support the cytolysis mechanism we have proposed and suggest controlling it by membrane curvature modulation. We have demonstrated it for fibroblast and keratinocyte by incorporating solutes instead of osmotic pressure to control CPP cytolysis 9,11 . Hoffmeister Effect of solute 12 is one of the known phenomena to control the hydration level of self-assembled amphiphiles. Chaotropic solutes (water structure breakers) such as sodium thiocyanate (NaSCN) and 1,3-butanediol (1,3-BG) make the membrane curvature positive. On the other hand, kosmotropic solutes (water structure makers) such as sucrose make the membrane curvature negative 14,15 . Chaotropic solutes, such as NaSCN and 1,3-BG enhanced cellular uptakes of FITC-R8 without interfering with  Here, we have confirmed these solute effects for FITC-R8 cytolysis for GUV as well.
The uptake of FITC-R8 to the GUV was measured by the same procedure as the osmotic pressure change, but inside is pure water instead of PBS and the outer media was changed to an aqueous solution of NaSCN, NaCl and Na 2 SO 4 at a concentration of 1700, 170 and 17 mM respectively. The translocation of FITC-R8 was c.a. 100 (µmol/ mol lipid) for NaSCN, c.a. 50 (µmol/mol lipid) for NaCl and c.a. 20 (µmol/mol lipid) for Na 2 SO 4 . The order of   In this case, the mean curvature of lamellar membrane changes to negative because both DOPE and DOPS have a smaller hydrophilic group (smaller a) than DOPC, resulting in SP increase. During this phase change, the transition of Mesh 2 can be formed which has hyperbolic surfaces with holes as catenoid saddle-splay negative Gaussian curvature as well. Such reverse curvature generation to V 2 and H 2 only happens through the Stalk structure 21 . The Stalk structure is a well-accepted model for cell fusion or fission processes and endocytosis (Fig. 7b).
Consequently, local and temporal positive curvature change is a crucial mechanism to make CPP penetrate directly into the cytosol (Fig. 6). Nature has systems to prevent fatal pore formation by clogging the pores with large membrane protein or creating local lipid domain (raft) (as shown in Supplementary Fig. S7) 22 . Local adsorption of CPP would be too quick for such a defense system to respond.
While we would not exclude the importance of the Stalk model to explain CPP's penetration, but this path would be generated by negative curvature to lead to the formation of endosome as discussed (Fig. 7). There were no endocytosis uptakes observed in the experiments reported here.

Conclusion
In conclusion, local and temporal positive curvature change seems crucial to promote the CPP cytolysis. With this knowledge, controlling the translocation of CPP by physical or chemical stimuli should become a promising way for both the development of advanced DDS and the prevention of CPP's invasion.
The mechanism proposed here could be expanded to other phenomena and applications as a general mechanism to control the bi-layer membrane's structural change. One possible example may be the permeation of cationic nanoparticles through the cell membrane since adsorption of cationic nanoparticles to the phosphate anion of planar and flexible lipid membrane would increase the cross-sectional area a. which increases the curvature (decrease SP) to generate local and temporal Mesh 1 structure. Utilization of CPP or any material and surrounding condition to induce positive curvature could prevent the formation of the Stalk intermediate. Although these assumptions must be experimentally proven, utilization of this proposed mechanism as a temporal and local curvature modulation would open up a novel way to combat membranous malfunctions.

Materials and methods
Materials. Dioleoyl-phosphatidylcholine (DOPC), palmitoyl-oleoyl-phosphatidylcholine (POPC), and eggyolk phosphatidylcholine (EPC lipids provided by NOF Co., Tokyo, Japan) were used for the preparation of GUV. Phosphate buffered saline (PBS, Wako Pure Chemical Co. Ltd., Osaka, Japan) was used as a solvent. Oligoarginine (Rx, x = 4, 6, 8), used for the phase behavior of DOPC under equilibrium, were synthesized in the Fmocsolid-phase manner at Imura laboratory. FITC-R8 (fluorescein isothiocyanate conjugated with γ-aminobutyric acid as a linker to N-terminus of R8, synthesized at Futaki laboratory) was used for penetration study as CPP. Rhodamine B (Rho, 98% pure; Kanto Chemical Co. Inc., Tokyo, Japan) was used as the leakage marker. Triton-X100 (TX100, Sigma-Aldrich Inc., Missouri, USA) was used for the destruction of liposomes to examine the penetration ratio of FITC-R8 and leakage ratio of Rho. www.nature.com/scientificreports/ Preparation of erythrocyte. The preparation method was conducted in the same manner previously reported 9 . 5 ml of preserved sheep blood (Kojin Bio Co. LTD., Saitama, Japan) was centrifuged (4 °C, 3000 rpm, 3 min) and washed with 5 ml of physiological saline and centrifuged (4 °C, 3000 rpm, 3 min). The same operation was repeated to obtain erythrocyte pellet as a precipitate. 400 μL of erythrocyte pellets was washed with 400 μL of 20% PBS (isotonic) and centrifuged. An erythrocyte pellet for the cytolysis assay was obtained after the repeated operation.

Preparation of GUV.
GUVs were prepared by the freeze-melting and natural swelling method 23 . Lipids dissolved in chloroform were placed into the vials, and then the solvent was evaporated. After being dried in vacuo for 16 h in the dark, lipid films were hydrated with 0.01 mg/mL of Rho/3 M KCl solution to give 20 mM lipid concentration. Lipid suspensions were sonicated for 30 min (CS-20, SHIBATA SCIENTIFIC TECHNOLOGY Ltd., Saitama, Japan), frozen using liquid nitrogen rapidly, and melted at room temperature. The suspension was vortexed for 1 min. This step (from freezing to vortexing) was repeated six times. It was transferred to a dialysis membrane (UC-8-32-25; Viskase Co. Ltd., Chicago, USA) and dialyzed against 20% PBS for at least two days at room temperature and, the dialysis solution was daily replaced by a new one. In this study, 20% PBS is defined as 'isotonic solution' and PBS solution at concentrations less than 20% as 'hypotonic solution' and at higher than 20% as 'hypertonic solution' .
Measurement of particle size distribution and microscopic observation of GUV. GUV's particle size distribution was measured with a dynamic light scattering measuring apparatus (DLS, Nicomp 380ZLS, Agilent Technologies, Tokyo, Japan) using an argon laser (532 nm). For the observations of GUV prepared differential interference contrast microscope (DICM, ECLIPSE E-600 Nikon, Japan) was used.
Small-angle X-ray scattering (SAXS) measurements. SAXS experiment was carried out using a SAX-Sess camera (Anton-Paar Co., Ltd., Graz, Austria) and a PW3830 X-ray generator (PANalytical Ltd., ALMELO, Netherlands) was operated at 40 kV and 50 mA. All samples were filled into a thin quartz capillary and set in a sample holder unit, which can control the temperature within 0.1 °C accuracy. (TCS120, Anton-Paar Co., Ltd., Graz, Austria). An imaging plate was used for recording the scattering data and read out by a Cyclone storage phosphor system (Perkin-Elmer Co., Ltd., Massachusetts, USA) to get scattering data.
Penetration assay of CPP into erythrocyte. The penetration assay of CPP into erythrocyte was conducted in the same manner previously reported 9,10 . To the 400 μL of erythrocyte pellet, 400 μL of PBS at concentrations of 98 mM (196 mOsm as hypotonic), 140 mM (280 mOsm as isotonic), and 182 mM (364 mOsm as hypertonic) were added for the evaluation of cytolysis by CPP. 20 μL of 40 μM FITC-R8 was added in the 1.40 * 108 cells of sheep erythrocyte. After incubation for 10 min at 37 °C, the solution was centrifuged (4 °C, 12,000 rpm, 10 min) and washed with 400 μL of isotonic PBS and centrifuged (4 °C, 12,000 rpm, 10 min). After the repeated washing process, the precipitate was mixed with 200 μL of 10% TX-100 isotonic PBS to solubilize the membrane. FITC-R8 penetrated to erythrocyte was determined by fluorescence spectrometry (Ex = 495 nm, Em = 520 nm). All measurements were used five samples (n = 5), and a significant difference was calculated by using the t-test (*α value was fixed at 0.001 and p < 0.001, **α value was fixed at 0.01 and p < 0.01). Throughout these experiment, no hemolysis was observed within the range of osmotic pressure tested, which suggests that membrane barrier for hemoglobin in the erythrocyte was kept intact 9 .
Penetration assay of CPP into GUV. Penetration assay of CPP into GUV was conducted in the same manner previously reported 23 . 600 μL of 20 mM GUV was centrifuged (4 °C, 10,500 rpm, 20 min) and washed three times with 20% PBS and 1200 μL of PBS at concentrations at 5, 10, 20, 30, and 40% (14, 28, 56, 84, and 112 mOsm) were added to precipitate and control the shape of GUV. The following experiments were done on separating five liposomal suspensions. That is, 200 μL of 10 mM GUV was mixed with 780 μL of each concentration of PBS and 20 μL of 100 μM FITC-R8. In this case, the molar ratio of lipid/peptide (L/P) was fixed at 1000. A total of 1 mL samples were incubated at 37 °C or 2 °C for 10 min in the water bath and centrifuged (4 °C, 10,500 rpm, 20 min) two times. The precipitate and 100 μL of supernatant were mixed with 200 μL of 10% TX-100 and 100% PBS to make the total volume 1 mL. The amount of penetrated FITC-R8 was analyzed using the fluorescence spectrometer (Ex = 490 nm, Em = 520 nm, RF5300PC; SHIMADZU Co., Kyoto, Japan). Similarly, Rho's amount entrapped was analyzed using the same instrument as for Rho (Ex = 555 nm, Em = 575 nm). The ratio of penetrated FITC-R8 and Rho leakage ratio was calculated using the following equations: In Eq. (1), P is the penetrated ratio of FITC-R8 ratio, A is the amount of penetrated FITC-R8 into GUV calculated using the standard curve, and U is the amount of FITC-R8 used in this experiment. In Eq. (2), L is the leaked Rho ratio, and I L-Rho and I NL-Rho are the fluorescence intensities of leaked and non-leaked Rho. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.